Hybrid Control And Information Fusion Study On Construction Cluster MAS

The intelligence, dynamic scheduling and online prediction of construction quality of construction clusters are inevitable tendency of the future construction machinery industry.These study will deal with interception, backlogging and separation of the asphalt mixture. The problems of machinery allocation irrationality, needless waste of the mixture, hardness to guarantee construction quality and labor intersity will also be solved. In this thesis, MAS (multi-agent systems) is used to build a distributed architecture for the construction clusters and it is realized on the platform of MAGE. This architecture thoroughly represents the autonomy of the single machine and the flexibility of the clusters. The choice of a rational dynamic scheduling is a puzzle in construction clusters. In this paper, dynamic switching trategy of hybrid control and optimizing arithmetic are applied in analyzing material supply and dynamic scheduling of construction clusters. Information fusion is a useful tool to online forcast construction quality and four different methods of forcasting are used to predict the construction quality real-time.The main research contents of this thesis are listed here:1) For the first time, the platform of MAGE is applied to build the distributed architecture for construction clusters. This model gives attention to distributed and centric architecture and it can add and reduce single machine flexibly. The actualization project of intelligent construction clusters is provided. It includes the structure of center control agent, IDSS, knowledge base, construction criterion of single machine, man-machine alternation and wireless communication.2) Information fusion based on time series NN, SVM, stochastic gradient, and MARS are studied to forcast the construction quality. The simulation results show that these methods are effective.3) Based on the coupling relation of supplication and requirement of the asphalt mixture in construction, the system thresholds are given and hierarchical model and MLD model of the system are built to analyze the...